#1 Defensemen and their Impact (based on Goals)

[weastern bias]

In general, it's pretty hard to play a guy 25+ minutes and shelter him.

Unless he plays nearly 2 minutes per PP

 

[aufheben]

This doesn't include information like Hedman playing with Stralman or McDonagh playing with Girardi though. Still interesting. McDonagh is starting to return to his 2013-14 form.

Same with McD.

He's unfortunately spent most of his career with statistically one of the worst defenseman in the NHL. McDonagh-Stralman were absolutely dominant the times they were together. Even when McDonagh has been on his off-side and paired with Staal, it's been better.

This season McDonagh-Girardi is 42.1CF%, while McDonagh is 55.2CF% away from Girardi, and Girardi is 43.3CF% away from McDonagh. It's absurd.

 

[NyQuil]

Unless he plays nearly 2 minutes per PP

Are you talking about Burns? He's still 8th in the league in ES icetime and first on the Sharks. (20:07)

Karlsson is 2nd in the league in ES icetime. (23:03)

Sheltered is a guy like Tom Preissing who put up a +42 for the Senators playing 14 minutes a game.

 

[bossram]

yes just this season

Appreciate the effort but a single (partial) season is just way too small a sample to get any accurate picture of these players.

We know that goal differential is heavily influenced by SH% and SV% fluctuations that are essentially out of player control for the most part. It would take an extremely large sample for the goal differentials to even out.

Ex: Kris Letang.

This season he has a net impact of -1.71 according to your metrics. His GF% is 39.6% (at ES) and a -10 overall differential. Definitely not good. BUT all of his Corsi, Fenwick, Scoring Chances and High Danger Scoring Change percentages are well above 50% and positive differential.

Why are his goal numbers awful? He has a 4.88% on-ice shooting percentage. Well below league average of ~8%. He's playing well, but when he's on the ice no one is managing to score. It's reminiscent of what happened to Edler a couple of years ago when everyone thought he sucked because his +/- was bad. Posters like me pointed out his on-ice SH% was 4% and lo and behold, he "turned it around" next season. He didn't. He was playing the same by the underlying numbers.

To illustrate the problem with the sample size, Letang's Goals For % is 53% since the Pens cup run. He's a positive impact player.

 

[JayKing]

Shows how underrated Subban & Hedman are and how scary good Ekblad is right now and how much more he can be

 

[HarrisonFord]

Appreciate the effort but a single (partial) season is just way too small a sample to get any accurate picture of these players.

We know that goal differential is heavily influenced by SH% and SV% fluctuations that are essentially out of player control for the most part. It would take an extremely large sample for the goal differentials to even out.

Ex: Kris Letang.

This season he has a net impact of -1.71 according to your metrics. His GF% is 39.6% (at ES) and a -10 overall differential. Definitely not good. BUT all of his Corsi, Fenwick, Scoring Chances and High Danger Scoring Change percentages are well above 50% and positive differential.

Why are his goal numbers awful? He has a 4.88% on-ice shooting percentage. Well below league average of ~8%. He's playing well, but when he's on the ice no one is managing to score. It's reminiscent of what happened to Edler a couple of years ago when everyone thought he sucked because his +/- was bad. Posters like me pointed out his on-ice SH% was 4% and lo and behold, he "turned it around" next season. He didn't. He was playing the same by the underlying numbers.

To illustrate the problem with the sample size, Letang's Goals For % is 53% since the Pens cup run. He's a positive impact player.

That's the thing though, and that's what makes the numbers so interesting to me. Each player has contributing factors for why their numbers are the way that they are.

 

[NyQuil]

I never mind looking at the numbers and then trying to explain them.

It's not the fault of the numbers.

There are usually valid reasons why numbers are the way they are.

The interpretation is as interesting as the figure.

 

[lomiller1]

It also does not account for on ice sv%/on ice sh%... they are not all luck but in many circumstances are heavily luck related, especially over a ~40-50 game sample.

Would be interesting to see both Corsi and Shot charts as well.

I mean, going off GF rel stats this season Luca Sbisa is elite... when he has some of the worst corsi rel stats in the league. He just has a horseshoe up his arse atm.

This was my thinking as well. Usage matters as well but usually not as much as people think because players end up facing similar usage regardless of what their coaches try. Eg, nearly all players face competition in the range of 49% - 51% CF%

 

[HarrisonFord]

I never mind looking at the numbers and then trying to explain them.

It's not the fault of the numbers.

There are usually valid reasons why numbers are the way they are.

The interpretation is as interesting as the figure.

NyQuil, you just get me dude.

 

[aufheben]

This was my thinking as well. Usage matters as well but usually not as much as people think because players end up facing similar usage regardless of what their coaches try. Eg, nearly all players face competition in the range of 49% - 51% CF%

Who they play with is probably more important than their competition.

 

[Nurmagomedov]

Player | GFW60 | GFWO60 | GAW60 | GAWO60 | NET60
H. Lindholm | 0.99 | 2.05 | 1.33 | 2.56 | -1.06- (-0.51) = -0.55

How does that NET60 part work? Specifically how does (GAWO60 - GFWO60) make sense? The player isn't on the ice in either case.

 

[Gentle Man]

Cant speak for all teams, but this chart just confirms what Ive watched all year in regards to Ekblad, ESPECIALLY starting from game 24 to date (when Campbell and Ekblad was split up).

 

[bossram]

That's the thing though, and that's what makes the numbers so interesting to me. Each player has contributing factors for why their numbers are the way that they are.

I never mind looking at the numbers and then trying to explain them.

It's not the fault of the numbers.

There are usually valid reasons why numbers are the way they are.

The interpretation is as interesting as the figure.

It always comes down to interpretation.

My point is there isn't much to interpret from partial, single season goal numbers because they are so volatile. They can't tell us anything really meaningful.

 

[HarrisonFord]

It always comes down to interpretation.

My point is there isn't much to interpret from partial, single season goal numbers because they are so volatile. They can't tell us anything really meaningful.

Well that depends on if you're talking about a player in the general sense (i.e. how good PK Subban is) and what his impact has been on the team this year

 

[Aceboogie]

Can you post the resource you used for this? Id like to recreate it with Klefbom and not Sekera. I was on stat.hockeyanalysis but I couldnt recreate anyones. Although good research. This is what HF needs more of, objective inquiry. So I commend this

 

[LaP]

Yeah, I think any d-man playing over 25 minutes shouldn't be tagged for matchups in any way since it's pretty clear he's out playing with and against practically everyone.

Yep when our 4th line scores a goal Subban often gets an assist on it

 

[HarrisonFord]

Can you post the resource you used for this? Id like to recreate it with Klefbom and not Sekera. I was on stat.hockeyanalysis but I couldnt recreate anyones. Although good research. This is what HF needs more of, objective inquiry. So I commend this

it was stats.hockeyanalysis.

The reason you might be having trouble recreating is because I used only the GF/GA stats with the starting goalie. So for Edmonton, I checked Talbot's numbers with and without Sekera

And much appreciated

 

[BigDaddyLurch]

...so Subban isn't as bad as many here would have us believe??...colour me surprised...

 

[Aceboogie]

I used Klefbom instead of Sekera because I believe hes our #1 (plays tougher comp, better D)

The majority of his time is spend with Nilsson and not Talbot. Which makes sense because he went down in December when Nilsson was solid and getting more starts (also Talbot was shakey 1st part of the season)

Using Nilsson he has 0.44 net (0.13 GF, -.31 GA) which ranks him around Suter at 6th best. Which is solid for him, hes been great this season

 

[Hagged]

Negative defensive impact is upside down. Consider Ristolainen. Starting on the d-zone against top6 lines means his teams GA/60 will inevitably be higher "with him". Corsi number as quality of competition is also bad stat crunching, shots created by first lines aren't equal to shots generated by shut-down/energy lines. Without Ristolainen, Buffalo would be even worse defensively, as his pairing is the only one that can match against opposition first lines on equal ground.

 

[Aceboogie]

How does that NET60 part work? Specifically how does (GAWO60 - GFWO60) make sense? The player isn't on the ice in either case.

Shows what the team can do without him on the ice. If team can prevent goals well on the ice without you and then when you come on the GA go way up, its likely because you arent good defensively and affecting that

Similar with GF. If your team cant score with you on the ice, and then you go on the ice and double GF, you are good offfensively

 

[Aceboogie]

Negative defensive impact is upside down. Consider Ristolainen. Starting on the d-zone against top6 lines means his teams GA/60 will inevitably be higher "with him". Corsi number as quality of competition is also bad stat crunching, shots created by first lines aren't equal to shots generated by shut-down/energy lines. Without Ristolainen, Buffalo would be even worse defensively, as his pairing is the only one that can match against opposition first lines on equal ground.

Using every teams #1 D ( in terms of quality of comp, minutes) then every D equals out, because every D in this comparison gets same disadvantage

 

[hersky77]

...so Subban isn't as bad as many here would have us believe??...colour me surprised...

Even in this drop from hell or whatever you want to call it, he is still producing and playing like a top 5 dman.

 

[tsujimoto74]

Just out of curiosity, who did you use for Buffalo's #1G in the comparison? Lehner is obviously the guy, but injuries have limited the number of games he's played in. Johnson and Ullmark split starts fairly evenly while he was out (w/ Johnson playing a bit more).

 

[HarrisonFord]

Negative defensive impact is upside down. Consider Ristolainen. Starting on the d-zone against top6 lines means his teams GA/60 will inevitably be higher "with him". Corsi number as quality of competition is also bad stat crunching, shots created by first lines aren't equal to shots generated by shut-down/energy lines. Without Ristolainen, Buffalo would be even worse defensively, as his pairing is the only one that can match against opposition first lines on equal ground.

Right, but that's why I'm not comparing #1D to #4D. So just like for Ristolainen, these other #1D are playing against top 6 lines. But Ristolainen is still new in the league, so you'd expect that to improve as he gets older